Onion Pollination as Affected by Different Levels of Pollinator Activity

Transcription

1 Utah State University All PIRU Publications Pollinating Insects Research Unit Onion Pollination as Affected by Different Levels of Pollinator Activity George E. Bohart Utah State University William P. Nye Utah State University L. R. Hawthorn Follow this and additional works at: Part of the Entomology Commons Recommended Citation Bohart, G. E., W. P. Nye, and L. R. Hawthorn Onion Pollination as Affected by Different Levels of Pollinator Activity. Utah Agr. Exp. Sta. Bull. 482, figs., tables. This Bulletin is brought to you for free and open access by the Pollinating Insects Research Unit at It has been accepted for inclusion in All PIRU Publications by an authorized administrator of For more information, please contact

2 ONION POLLINATION - AS AFFECTED BY DIFFERENT LEVELS 'O F POLLINA1:0R ACTIV G. E. Bohart, W. P. Nye and L. R. Hawthorn Bulletin October, 1970 Utah Agricultural Experiment Utan State University

3 G. E. BOHART and W. P. NYE are with the Entomology Research Division, Agricultural Research Service, USDA, logan, Utah R. HAWTHORN was formerly with the Crops Research Division, Agricultural Research Service, USDA, Logan. He is presently at 113 Wiggins Street, West Lafayette, Indiana / -

4 ACKNOWLEDGl\t!ENTS Weare indebted to Mr. John R. Pease now at the University of Utah Medical School in Salt Lake City; Dr. Charles 1. Graham, now at the U. S. Army Biological Laboratories at Fort Dietrick, Maryland; and Mr. Ross A. Nielsen, now at the Southeastern Bee Research Laboratory at Baton Rouge, Louisiana, for making many field counts and measurements during the course of these studies. Dr. Lynn Kerr, now at the Charter Seed Company at Twin Falls, Idaho provided most of the plot management and could be counted on when needed. Many of the insects collected on the plots were identified by specialists, including the following: Hymenoptera Aculeata (except Apoidea) - Apoidea - G. E. Bohart Braconidae - P. March Chalcidiodta - B. D. Burks R. M. Bohart Diptera Bombiliidae - J. C. Hall Stratiomyidae - W. J. Hanson Syrphidae - H. S. Telford Therevidae - 1. Knutson Muscoidea, Chloropidae - C. Sabrosky Otitidae, Tephritidae - G. Steyskal 1

5 CONTENTS ACKNOWLEDGEMENTS INTRODUCTION 1958 Methods and Materials Conditions Affecting the Experiment. Results 1959 Methods and Materials Conditions Affecting the Experiment. Results 1960 Methods and Materials Conditions Affecting the Experiment. Results 1961 Methods and Materials Conditions Affecting the Experiment. Results DISCUSSION AND CONCLUSIONS SUMMARY LITERATURE CITED APPENDIX A

6 ONION POIJJNATION AS AFFECTED BY DIFFERENT LEVELS OF POIJJNATOR ACTMTY G. E. Bohart, W. P. Nye, and L. R. Hawthorn According to Jones (1937), the pollination of onion (Allium cepa Linnaeus) is effected mainly by insects that visit the nectaries at the base of the three inner stamens. Muller (1833) and Jones (1937) stated that cross-pollination is the rule, although self-pollination between flowers of the same umbel is undoubtedly of frequent occurrence. Muller (1883) and Hayward (1938) reported that the flowers are imperfectly proterandrousthe anthers of the inner whorl of stamens dehisce before those of the outer. Both whorls dehisce before the stigma is receptive. Moll (1953) stated that the normal receptive period of the stigma is 3 days. Several investigators have noted that onion flowers are visited freely by many kinds of insects, including honey bees. (Apis melli/era Linnaeus) (Muller, 1883; Jones and Emsweller, 1934; Shaw et. at. 1935; Kordakova, 1956). For this reason, it has often been taken for granted that pollination is not a limiting factor in onion seed production in spite of the tremendous variation known to occur in yields from different fields, seasons, and cultural practices (Morse, 1923; Hawthorn, 1951). The recent development of hybrid seed grown on male-sterile inbreds pollinated by male-fertile inbreds has intensified the pollination problem. Pollinating insects are required to disseminate pollen over 6 to 12 or more rows from two pollen parent rows Complicating problems are the supposed relative unattractiveness of malesterile plants to pollen-collecting insects and the possibility that their receptive period will not coincide with the pollen dehiscence period of the pollen parents. Another problem, and one particular to the Treasure Valley of Idaho and Oregon (the most important onion seed producing area in the U. S.) has been the great increase in alfalfa seed production, resulting in thousands of acres of alfalfa bloom which drain off populations of honey bees and other pollinators. Finally, increased use of insecticides has reduced the populations of pollinators. In Treasure Valley, the widespread use of Dylox (sometimes called Dipterex) for alfalfa insect control has probably had a particularly disastrous impact on dipterous pollinators. 3

7 In 1958, we started a 4-year program to learn more about the pollination factor as it affects onion seed production. In the following discussion, the methods, conditions, and results for each year will be treated separately, and a general discussion will follow. Handling of the plots and methods of measurement were the same from year to year unless otherwise stated Methods and Materials Location White farm, northeast bench, Logan, Utah. Treatments The three treatments (randomized within the row of plots) were as follows: (1) Plots caged to admit only tiny insects; (2) plots without cages (open pollination); and (3) plots caged to enclose a small (four-frame) colony of honey bees. The cages (Pedersen et ai., 1950) were 21.5 feet long, 11 feet wide, 6 feet high, and covered with 12-mesh clear lumite screen. As soon as flowering began in late June, the cages were erected and the hives of bees installed. Plot layout In 1958, there were 12 plots that measured 10 x 20 feet arranged in a single row. Each plot contained five rows, each planted with 40 onion bulbs. The rows were 2 feet apart, and the plants were spaced ab()ut 6 inches apart in the rows. The center row was plante:! with U B, a male-fertile inbred line. The other four rows were planted with the following three male-sterile inbreds: B-1900A, B-2147A, and B-2267A. Two rows of the B-2267 A were planted in each plot and the positions of all the male-sterile rows were randomized. More than one male-sterile line was used merely because we were unable to obtain enough bulbs of a single line. When the bulbs were obtained, we anticipated that male sterility would be complete, but at least a small amount of pollen was present in all of the lines. Irrigation and insect control The plots were irrigated once each week throughout the growing season. 4

8 The onion maggot [(Hylemya antiqua (Meigen)] was controlled by the application of Dieldrin in the furrow at the time the bulbs were plantl~d. Onion thrips (Thrips tabaci Lindeman) were controlled with a foliage application of 10-percent DDT dust just prior to spfitting of the sheath on the umbel. No insecticides were applied after blossoming started. Eriophyid mites were not recognized as a potential problem, and, hence, no attempt was made to determine their presence or to control them. Measurements 1. In 1957, in the course of a series of experiments on carrot pollination (Hawthorn et ai., 1960), the effect of the cages on air movement, air temperature, and light intensity at the flower level, and on temperature at the soil mrface, was studied in detail. In 1958, we made another series of readings in our onion pollination studies to find the effect of the cages on relative humidity and soil surface temperature. 2. Nectar sugar concentration in the flowers was measured on July 18 on all plots (except those excluding large insects) by reading the sugar concentration of the nectar in the honey stomachs of bees taken directly from the flowers. A Bausch Lomb'*' low rang.e (0-60 percent) hand refractometer was used for this purpose. 3. Flowering was measured 12 times throughout the season on the basis of the number of flower heads on which at least half of the flowers had opened. The duration of bloom was not measured. 4. On the mornings and afternoons of July 18 and 21, the relative numbers of nectar-collecting and pollen-collecting honey bees were measured in the cages with bees enclosed. 5. Counts of each species of pollinator were made on each row in the open plots. Four counts were made on each of 12 days th:o:.i.shout the season. The ob~erver 's route from plot to plot and row to row was changed for each observation. Tiny flies were difficult to see and their p~pulations were undoubtedly understated by the counts. Populations of tiny flies in the cages excluding bees were cnly evaluated in general terms. 6. A pollination efficiency rating for each pollinator was assigned by estimating the quantity of loose pollen grains fro.:n a representative sample and combining this factor with the size, hairiness, and a:tivity pattern of the insect. This type of rating, used previously br carrot pollinators (Bohart and Nye, 1960), is admittedly subjective, but whea it is multiplied by the populations to give a pollination index, it should give a truer 5

9 picture than population figures alone. In the case of honey bees, separate ratings were given to nectar and pollen collectors, and the overall rating for the species was based upon the proportion of the two activity patterns taking place. 7. The seed heads were dried in September and threshed on a small rubbing board. The yield data in 1958 and subsequent years were subjected to analysis of variance. The significance of mean differences was determined by the application of Duncan's (1955) multiple-range test. Soil conditions Conditions Affecting the Experiment The soil was classified as Timpanogos silt loam. The moisture level was adequate for the row spacing throughout the growing season. Plants Some of the bulbs were decayed and, as a result, a number of plants failed to develop and others were only" partially productive. Since spacing in the rows was wide, it was possible to correct the data for missing plants without being concerned about increased yield on plants adjacent to the missing ones. Harmful insects The populations of onion maggot and onion thrips were held at a low level by the insecticide treatments used. Mirid bugs (Lygus spp. and Orthops scutellatus Utler) were present in the bloom stage but did not exceed one or two per flower head. Cage effect The principal effect of the cages was to reduce light intensity by about 39 percent and air movement by abc:.lt 45 percent (table 1). Th~ temperature and humidity in the cages and in the open were nearly the same, apparently as a result of the offsetting effe~ts of shading and reduced air movement. The temperature at the soil surface was somewhat lower in the cages in both the carrot and onion studies, perhaps because air movement differences were less pronounced at this level. 6

10 Table 1. Air movement, relative humidity, air temperature, an d light intensity at flower height, and soil temperature at the ground level in the open and caged plots. Carrot and onio n pollination experiments, Logan. Utah, 1957 and 1958, respectively Treatment Year Air movement* ft. per min. Relative humidity** (percent) Air temperature (of) Light intensity* * * (meter readings) Soil temperature (of) " Open Caged * Effect of cage measured in a closed room in which a fan created an artificial breeze of 500 feet per minute. Measured with Electric Hygrometer indicator, American Instrument Co., Inc., Silver Springs, Maryland. Measured with Weston Master light meter (model I) directed toward a neutral gray (18% reflectance) card.

11 Weather In July of 1958, when flowering took place, temperatures were unusually low for the Logan area and precipitation was moderate (table 2.) Nectar sugar concentration The sugar in the nectar of the plants in the open plots was nearly twice as concentrated as that in the plants in the cages with honey bees (table 3). Greater air movement in the open plots probably increased nectar evaporation to cause this result. More rapid removal of nectar in the bee cages, with consequent less time for evaporation, was probably also involved (Nye and Pedresen, 1962). In support of this, we noted that the cage with the weakest colony of bees had flowers with the highest nectar sugar concentration (29 percent). Progress of bloom The cumulative progress of opening blossoms on the open plots is shown in figure 1. The point of decline in total bloom appeared to occur on or about July 15 when the cumulative curve flattened out. There was little difference among inbred lines in the progress of bloom. Table 2. Temperature (OF) and precipitation data for the Logan, Utah area in July Temperature No. of days Total Average Average 90 0 F or precipitation Year maximum minimum Highest above (inches) * * 7 days over 100 F. Table 3. Sugar concentration in the nectar" of onion flowers on open plots and plots with honey bees enclosed, onion pollination experiment, Logan, Utah, 1958 Treatment Open plots 44 Plots with honey bees enclosed 21 P_ercent sugar concentration By replication Average * Nectar taken from the honey stomachs of honey bees collected from the onion flowers. Each figure represents 16 samples (four bees collected four times during the season)

12 All1naecta \0 40 I ~ 30 LA! 5!.!3 t 8 ~ ~ I I I I,,' "-/ ~' I ~ --.~ / II / /T./ ~ Honey bees Number of onion heads per plot \, ----,...---~- \ "-,, \,,,, " ", "... '\,,~ ~ 350 of Qj '"' 0. CIl 200 -g Qj.d 1' Qj '"' J 100 1l ::j z Days Figure 1. Seasonal fluctuations of honey bees and all insect populations compared with accumulative total of open flower heads, 1958.

13 / Pollination population-tiny insects only In the cages of treatment 1, the only potential pollinators seen were small Diptera, including ceratopogonids, sciarids, cecidomyiids, chloropids, milichiids, and a few small spyphids [(mostly Syritta pipiens (Linnaeus)] The populations of tiny flies in the cages were small and few of the individuals examined carried pollen grains. Pollination population-open plots (number per 100 feet of row) Throughout the season, the average number of insects per 100 feet of row per observation was 32. The corresponding pollination efficiency rating was 52. The most abundant order of insects was Diptera (average, 22). Next was Hymenoptera (11). All other insects (principally Lepidoptera) were relatively scare (0.1). In terms of pollination efficiency, the above groups averaged 27, 15, and 0.12, respectively (tables 4, 5). The most abundant insect was a syrphid fly (Syritta pipiens (9» and next was a chloropid fly [T hattmatomyia glabra (Meigen) (6)]. Both of these were ineffective pollinators and together they contributed only 10 percent to the total efficiency rating. Drone flies [Eristalis tenax (Linneaus) and E. brousii Williston (4.6)] were next in abundance among the flies. but because of their greater effectiveness as pollinators, they contributed almost half of the total efficiency rating for all insects. Among Hymenoptera, honey bees were the most abundant, but because most of them were nectar collectors, their contribution to the total efficiency rating (about 23%) was less than that of the drone flies. A species of sweat bee, Halictus farinosus Smith, was only one-fifth as abundant as the honey bee but contributed about half as much ( 11 % ) to the total efficiency rating. The only other insect species contributing to a marked degree (about 5%) was a sand wasp, Bembix amoena Handlirsch. The many other insect visitors assumed importance only in the aggregate. Pollinator populations tended to follow the curve of flower development. The plateau of high population from July 9 to 21 coincided with the period of greatest bloom. By July 28, bloo:n had deteriorated to a point lower than at the beginning of the counts but pollinators were still moderately abundant, perhaps because they were accustomed to visiting the plots. Population curves of some species of pollinators did not follow the curve. It can be assumed that these were strongly influenced by their own 10

14 seasonal population curves, or perhaps by the flowering periods of more attractive blooms within their flight range. For example, Hatietus!arinosus gradually increased its visitation to the plots until July 25, whereas Bembix amoena was most abundant on July 9. Variation in the population curves of different species resulted in a broad plateau of high insect population. (See appendix A for complete list of insect visitors to onion plots.) Pollinator population-caged plots with honey bees Unfortunately, accurate measurements of bee populations in the cages were not taken. The average number present during the majority of the blooming period was roughly estimated to be between 10 and 200 per plot, which, at slightly less than one bee per flowering umbel, was much higher than all insects combined in the open plots. In one replicate, the population was at least one-third lower than in the others. Table 6 shows the relative numbers of pollen and nectar-collecting honey bees in the caged plots with honey bees. There was a slight, but probably nonsignificant, increase in the percentage of nectar collectors throughout the day (85% to 93%). Seed yields R-esults In 1958, the overall seed yields from the enclosed plots with bees was 171 pounds per acre compared with 355 in the open plots and no seed in the plots with only tiny insects (table 7). Yields from the pollen parent were not significantly different between the plots with bees enclosed and the open plots, but yields from the male-sterile inbred lines were much higher in the open plots. Relative yields from the male-sterile lines were entirely different in the open and in the caged plots. Line B-2149A gave the highest yield in the cages but the lowest in the open plots. 11

15 Table 4. Average populations of insects per 100 feet of row on open onion Species of pollinators Apis mellifera Halictus farinosus Other bees All bees H Bembix amoena Vespula spp Other wasps All wasps All Hymenoptera Thaumatomyia glabra Syritta pipiens Eristalis tenax Other Diptera All Diptera All Lepidoptera All insects Table 5. Average population X efficiency of insects per 100 feet of row Species of pollinators Apis mellifera Halictus farinosus Other bees All bees Bembix amoena Vespula spp Other wasps All wasps Ail Hymenoptera Thaumatomyia glaljra Syritta pipiens Eristalis tenax Other Diptera All Diptera All Lepidoptera All insects

16 plots, onion pollination experiment, Logan, Utah, July Average on open plots, onion pollination experiment, Logan, Utah, 1958 July Average "

17 Table 6. Percentages* of nectar and pollen-collecting honey bees on the plots caged with honey bees, onion pollination experiment, Logan, Utah, July 21, 1958 Time of day Nector collectors Pollen collectors 8-10 am am pm pm 93 7 Average * Averages of 190 bees examined on the four replications at each time of day Table 7. Onion seed yields in pounds per acre by inbred line and pollination level, onion pollination experiment, Logan, Utah, 1958.Inbred lines Open plots Bees caged B-1900A B-2149A B-2267A B-2267A Pollen parent Average 423 A* 310 BC 387 AB 387 AB 270 C CO C 170 * Means not followed by the same letter are significantly different at the 1-percent level. Location 1959 Methods and Materials Greenville farm, North Logan, Utah Treatments In 1959, when it was apparent that plots caged to admit only tiny insects would not produce seed, we established instead plots covered with cages (modified-open) designed to allow free passage of pollinators. This was done primarily to determine whether differences in yields between open and caged plots might be associated with cage effect rather than pollinator activities. The three treatments (randomized within the row of plots) were as follows: (1) Plots without cages (open pollination); (2) plots provided with cages partially rolled up on the sides and open at the upper corners to allow free passage of pollinators (modified-open); and (3) plots caged to enclose a small (eight-frame) colony of honey bees. 14

18 Plot layout The 12 plots were randomized in two rows. The center row of each plot was planted with U B, a male-fertile inbred line. The other four rows were planted with the following four male-sterile inbreds: B-2264A, B-2267A, B-2190A, and U A. Only the male-sterile inbred B-2267 A was the same used in the 1958 study. The other sterile lines used had no counterpart in As in 1958, male sterility was not complete; a small amount of pollen was present in all of the lines. Measurements 1. Progress of bloom on each row was measured on July 13 by counting the total number of heads and the number not yet open. On July 23, the percent of the bloom was estimated on each of 10 heads at the north end of each row on the first three replications. 2. Populations and pollination efficiencies of insect VISItors were measured on mornings and afternoons of July 10 and July 13, and mornings only of July 16 and August The population of honey bees in the caged plots with bees was roughly estimated but not accurately counted. Soil and soil moisture Conditions Affecting the Experiment The soil was classified as Millville silt loam. The moisture level was adequate for the row spacing throughout the growing season, and cultivation was uniform in the nursery. Plants The bulbs were in good condition and the stands were excellent. liarnnful insects Populations of onion maggot and onion thrips were held at a low level by the insecticide treatments used. Mirid bugs (Lygus spp.) were present in the bloom stage, but they did not exceed one or two per flower head. 15

19 Weather In July 1959 when flowering took place, temperatures were high and precipitation was below normal, less than one-fourth that of 1958 (table 3). Progress of bloom The cumulative progress of opening blossoms on all plots was similar to that shown in figure 1 for The point of decline in the total bloom was not accurately determined, but it appeared to occur shortly before July 23. Pollination population-open plots Insect populations per 100 feet of row per observation were much higher (128.5) than in 1958 (29.5) in spite of the fact that the number of species observed in the regular counts was smaller. The corresponding pollination efficiency ratings were and The smaller number of species observed in 1959 was in part accounted for by the smaller number of counts. The fact that most of the counts in 1959 were taken during 4 the peak bloom accounts to a slight extent for the disparity between average populations for the 2 years, but nearly as strong a difference is apparent when only the days of peak population are compared. The most abundant order of insects in 1959 was Hymenoptera (average 108). Next, about one-fifth as abundant, was Diptera (18). All other insects, principally Lepidoptera, were relatively scarce (2.7). In terms of pollination efficiency, the above groups averaged 239.8, 34.1, and 2.7, respectively (table 8). Honey bees were more than twice as abundant as all other visitors combined. This contrasts sharply with the previous year when honey bees were only about one-fourth as abundant as all other insects combined. Approximately the same relationships held for pollination indices as for pollinators. In other words, honey bees appeared to be about average in efficiency both years. Based on accumulation of loose pollen grains, Halictus farinosus and Andrena prunorum were the most efficient pollinators. Pollen-collecting honey bees were next, followed by two species of drone flies (Eristalis). Pollination population-modified-open plots The populations in the modified-open plots were slightly lower (average, 111 per plot) than in the open plots (average, 128) on all counting 16

20 days but one (table 9). The difference was nearly all accounted for by the smaller honey bee populations in the modified open plots. Other species of pollinators did not appear to be affected by the presence of the rolled-up cages. Honey bees and solitary bees reached their peak populations on July 13 (table 8), but Diptera reached a sharp peak on July 16. On August 7, when little bloom remained, wasps (primarily Chlorion ichneumonium) and butterflies were much more abundant than they had been during peak bloom (table 8). Trends according to date were much the same on the open plots as on the modified-open plots (table 9). Most of the insects were more abundant during the afternoon than the morning counts (table 10. However, drone flies (Eristalis spp.) were more abundant in the morning, just as they were on carrot plots in previous years. The relative attractiveness, based on pollination indices, of the inbred lines on the open and modified -open plots is shown in table 11. On both types of plots, the order of attractiveness was the same. The pollen parent ranked fourth among the five lines represented. Pollinator populations - caged plots with honey bees Honey bee populations were high. On July 16 the population was estimated at 300 to 100 feet of row. Nectar collectors outnumbered pollen collectors on this occasion by about four to one. Seed yields Results Yields were considerably higher in 1959 (table 11) than in 1958 on the open plots (567 pounds per acre, compared with 356). From the analysis of variance (table 11), it indicates that yields from the modified-open plots were significantly higher than those from the open plots with bees enclosed. Table 11 shows that yields from the inbred male-sterile lines were significantly different, and the regression shown in figure 2 indicates that (1) yields benefited from increased pollinator activity within the ranges encountered on the open and modified-open plots, and (2) line differences in seed yield were at least partially accounted for by differences in attractiveness to insects. 17

21 Table 8. Average population and their pollination efficiency of insects per 1959 July 10 July 13 Pollinator No. Index No. Index Apis mellifera Halictus farinosus Andrena prunorum Other bees All bees Bembix amoena Chlorion ichneumonium Vespula spp Other wasps All wasps All Hymenoptera Syritta pipiens Eristalis tenax Other Diptera All Diptera All Lepidoptera All insects ~ Table 9. Average population and their pollination efficiency of insects per Utah, 1959 July 10 July 13 Pollinator No. Index No. Index Apis mellifera Halictus farinosus Andrena prunorum Other bees.5.5 All bees Bembix amoena Chlorion ichneumonium Vespula spp. Other wasps All wasps All Hymenoptera Syritta pipiens Eristalis tenax Other Diptera All Diptera All Lepidoptera All insects

22 100 feet of row on open plots, onion pollination experiment, Logan, Utah, July 16 August 7 Average No Index No. Index No. Index ~ feet of row on modified-open plots, onion pollination experiment, Logan, July 16 August 7 Average No. Index No. Index No. Index

23 506 ~ o ~ o.. Q) ~ o S Q. "C Q) Q) III... o III 200 S f c Pollination index 150 r b Y.2.294X+8.4 F Figure 2. Regression of seed yield on insect pollination index, 1959 (Each row in the open and modified-open plots is represented 'by' a dot.) I\J.J:,.. "0 3 (J'l VJVJ VJ... tv 0...L o CTlCTl CTlI'V ~ 0 CTl VJ...L OJ 0 VJ... 0 <o(j'l :""'0... I'V 000 CTl VJO 0,0 ~j~!l) 3 ~ N 0 (J'l VJ CD -I!l) r:::r CD 9 :J (J) CD ~ z?i 3 (;;. < CD ;::::;: tv 01 =» 0 -'0 (j) ::l I CD _. ()) Q.; en I!l) (J'lQ CDI I:J 90) >< IQ. ~ 0 I ~ CTl o 0 ()) (J'l i:;' z 1"0 o I 12-.1_ ':::: I» ::E: :J ::::!.I»,!l) g=6- _ en 2.':J ::lj eel go en en I 3 >< I 6- CD (J) -0 CD ~I"O I ~... :r>o ~Io... CD... CD CD ::l C::l... Q. 3 I» 1 0 CD I 1- >< I ~ I~

24 by time of day. Onion pollination experiment, Logan, Utah, July 10, 13, Bembix Chlorion Syritta Eristalis Total amoena ichneumonium pipiens tenax No. Index No. Index No. Index No. Index No. Index Table 11. Seed yields in pounds per acre by inbred line level, onion pollination experiment, Logan, Utah, and pollination Onion inbred Modified- Bees line open open enclosed Male-steriles B-2264A B-2267A B-2190A U A Pollen parent U B Average 567 B* 824A 507 B Average 991 A 677 A 378 C 416 C 671 B 633 * Means followed by same letter are not significantly different at 1-percent level Methods and Materials Location Greenville farm, North Logan, Utah. Treatments The four treatments, randomized within two rows of plots, were as follows: (1) Plots without cages (open pollination); (2) modified-open plots; (3) plots caged to enclose a small (four-frame) colony of honey bees; and (4) plots caged without insects for 2 days and provided with a small honey bee colony the third day. Plot layout There were 16 plots composed of the four above-described treatments replicated four times. The center row of each plot was planted with 21

25 inbred male-fertile pollen parents-u b. The other four rows were planted with the following four male-sterile inbreds: B-2267 A, B-2264A, U A, and B-5546A. Only the male-sterile inbred lines B-2264A and B-2267 A were used in the 1959 study. The pollen parent and the other two male-sterile lines had no counterpart in the previous studies. As in previous years, male sterility was not complete; a small amount of pollen was present in all of the lines. Measurements 1. Information about air and soil temperature, relative humidity) and light intensity was gathered as follows: (1) air temperature at flow'-;r height was measured by using a chemical mercury-filled thermometer, shaded to keep direct sunlight from the bulb. (2) The soil surface temperature was measured by a chemical mercury-filled thermometer with the column section mounted on a board and the free bulb held in contact with the soil by means of a spike connected to the board and thrust into the soil. (3) The percent relative humidity was measured in the center of each plot with an electric humidity and sensing element placed 6 inches above the ground surface. (4) The foot candles of light were measured with a photo light meter directed toward #a neutral gray card. 2. Percent of sugar in the nectar. 3. Percent of flowers with pollen on the stigmas from each row during peak bloom. 4. Percent of blasted (straw-colored) florets in the center of the flower beads. 5. Counts of insect visitors on 12 days spaced throughout the flowering season. Percentage of pollen-collecting honey bees was measured in all plots. 6. Seed yields. Conditions Affecting the Experiment Plants, soil, soil moisture Equivalent to those in Harmful insects Insect populations were held at a low level. count recorded was 0.6 per umbel. The highest lygus bug 22

26 Weather The weather was clear and unusually warm, with a number of days surpassing 100 F. Air temperature at flower height (table 12) The average of four maximum temperature readings in the open plots was 94.5 F., and in the modified open plots it was 95.5 F. The average temperature in both treatments with bees enclosed was 96.9 F. The analysis of variance indicates that there was no significant difference between treatments or replications. However, the increased temperature in the bee cages may have been more drastic during the 100 F. plus periods. This may have been enough to kill many of the florets and reduce the seed yields. Surface soil temperature The surface temperatures by treatment and plot are summarized in table 12. The analysis of variance shows that treatments were significantly different at the I-percent level. The modified-open plots and those caged continuously wjth bees had the older, darker screen, whereas those with bees caged intermittently had new clear screen, which allowed more light penetration. The lowest surface soil temperatures were recorded in the modified-open plots. Relative humidity Table 12 shows the relative humidity for each treatment. The analy StS of variance shows that the replications, but not the treatments, were Table 12. Nectar sugar temperature by treatment. Treatment Open plots Modified open plots Honey bees continuously Honey bees intermittently (1 day in 3) concentration, air Onion pollination Sugar nectar* (percent) 32 B* *. * 35 B 26 C 45 A temperature, and soil surfaae experiment, Logan, Utah, 1960 Soil Relative Air temp surface temp humidity (0F)** (OF)*** (percent) 94.5 A A 16.5 A 95.5 A B 15.8 A 96.8 A B 14.8 A 96.8 A A 16.6 A * Readings taken from four honey stomachs of honey bees collected from. each plot. * * Four readings taken at flower height from each plot. * * * Four readings taken 3 inches from plant base in center row of each plot. * * * * Means in a column not followed by the same letter are significantly different at the 1-percent level. 23

27 significantly different at the 5-percent level. The difference between replications probably resulted from the presence of slightly more soil moisture in some plots than others. Blasting of florets A number of flower heads on all plots were observed with a circular patch of straw-colored (blasted) florets in their centers. The patches varied in size from a few central florets to areas an inch across. Blasted florets were observed in previous years but only in inconsequential numbers. In 1960, they were more abundant in some lines (especially B-5546A) than others, but the differences between treatments were much greater. The plots enclosing bees continuously had 40 percent of the florets blasted; those with bees 1 day in 3 had 23.8 percent, the open plots 2.9 percent; and the modified-open, 1.6 percent. Seed was not produced in blasted florets. The percentages of blasted florets are summarized in table 13. Nectar sugar concentration The data for sugar concentration ip the nectar of the onion flowers are summarized in table 12. The analysis of variance based on the total of four nectar readings for each plot was significant at the I-percent level for treatments and not for replications. There was no difference between open and modified-open plots. The sugar in the nectar of the plants in the plots caged with bees 1 day in 3 was nearly twice as concentrated as that in the plants continuously caged with bees. Apparently, the bee visitation 1 day in 3 allowed evaporation to occur in the nectaries. The amount of pollen on the stigmas Although it was difficult to select stigmas in the same development condition in the different inbred lines, the percent of stigmas observed Table 13. Percent of florets blasted in the center of onion umbel by pollination level and inbred line, onion pollination experiment, Logan, Utah, 1960 Inbred lines B-2264A B-2267A B-5546A U A U B Average Open Modified- Bees caged Bees caged plots open continuously 1 day in C* 1.4 C 30.2 A 36.4 A 2.9 C 1.6 C 40.0 A 23.8 AB 8.1 BC 1.0 C 20.0 B 21.9 B 0.1 C 0.1 C 3.1 C 3.2 C 0.0 C 0.0 C 0.24 C 1.5 C Average * Means not followed by the same letter are significantly different at the 1-percent level. 24

28 with pollen was nearly te same for all treatments [bees caged continuously (38), bees 1 day in 3 (33), and modified-open plots (37)J. Pollination populations-open plots The overall figure for insect visitors per 100 feet of row (595) was about five times as high as in 1959, but the corresponding figure for pollination index (293) was only about twice as high (table 14). A high percentage of the insects in 1959 were tiny flies with a low pollination efficiency. Since the efficiency rating assigned them (0.1) was probably too high, the pollination efficiency rating for all insects was probably also too high. Honey bees were more abundant on the open plots than in previous years (159 per 100 feet compared with 92 in 1959 and 59 in 1958). However, only 3 percent of them were pollen collectors. The population of bees other than honey bees (7.4 per 100 feet) was comparable to that in 1959, but wasps (20 per 100 feet) were three times as abundant. Differences in attractiveness among inbred grown lines were extreme. For example, B-2267 A and B-2264A each had 850 visitors per 100 feet in the open plots. This was three times as many.as on the pollen rows (250) and nine times as many as on B-5 546A. Figure 3. Onion pollination, open plot,

29 Table 14. Utah, 1960 Numbers and poilination indi.:es of various kinds of pollinators in B-2267A B-2264A Pollinator No. Index No. Index Apis mellifera (pollen) Apis mellifera (nectar) Other bees All wasps Large Diptera SyriHa pipiens Tiny Diptera Other insects All insects Insects less honey bees Pollination population-modified -open plots Population trends of the various insects were similar to those in the open plots, but the figures averaged only about half as high (table 15). The overall figure was 293 per 100 feet compared to 595 in the open plots. The relative order of attractiveness of the lines was also the same in both treatments. Pollination populations-continuously enclosed honey bees The plots enclosing honey bees continuously had insect vlsltation levels comparable to those in the open plots, but the pollination indices were much higher because of the high proportion of tiny flies in the open plots. Pollination population-honey bees enclosed 1 of 3 days Visitation levels on the plots enclosing bees 1 day of 3 were only onefifth as high, even on the days when the bees were present, as on those with honey bees continuously. Analysis of the data for insect populations (table 16) shows that all treatments were significantly different at the I-percent level, the order being as follows: Plots enclosing bees continuously, open plots, modified open plots, and plots with bees enclosed 1 day of 3. The same order held for pollination indices. 26

30 the open plots (per 100 feet of row), onion pollination experiment, Logan, Inbred line U B (pollen parent) U A B-5546A Average No. Index No. Index No. Index No. Index Table 15. Numbers of various kinds of pollinators per 100 feet of row in the modified-open plots, onion pollination experiment, Logan, Utah, 1960 Pollinator Apis mellilera (pollen) Apis mellifera (nectar) Other bees All wasps Large Diptera Syritta pipiens Tiny Diptera Other insects All insects A ~ Insects less honey bees 506 B-2264A No Inbred line U No U A No A ~Average Table 16. Insect populations and pollination indices by treatment, onion pollination experiment, Logan, Utah, 1960 Treatment Honey bees continuously Honey bees 1 day of 3* *,. Open Modified-open Insect visitors per 100 ft. of row* A ** D B 292.6C Pollination index per 100 ft. of row A D B C * Four replications each counted on 15 separate days. * * Different letters signify differences significant at the 1-percent level. * * * Figures refer to coun:s made on the day bees were in the caged plots. Visitation the other 2 days was O. 27

31 Results Seed yields Seed yields in pounds per acre by treatment and inbred lines are summarized in table 17. The highest yields were obtained in the modified-open plots (389 pounds per acre). The open plots were next (309), the plots of bees caged 1 day of 3 were third 228), and the plots with bees caged continuously (162) were lowest. The analysis of variance shows a significant difference between all treatments at the I-percent level and no difference between replications. Table 17. Onion seed levels In pounds per acre by imbred line and pollination levels, onion pollination experiment, Logan, Utah, 1960 Open Modified- Bees caged Bees caged Inbred line plots open continuously 1 day in 3 B-2264A 771 AB* 838 A 379 CDE 538 BCD B-2267A 364 CDEF 585 BC 114 G 196 EF B-5546A 78 G 128 F 62 G 73 G U A 119 G 174 EF 105 G 157 EF U B 211 EF 219 DEF 149 F 176 EF Average * Means not followed by the same letter are significantly different at the 1 percent level. Figure 4. Onion pollination, modified-open plot,

32 Yield differences between inbred lines were also significant, and the same trend of differences occurred in all treatments. The order of attractiveness to insects of the various inbred lines (based on the open and modified-open plots) coincided with the order of seed yields for every treatment except the one with honey bees continuously caged. In this treatment, the yields of two lines were reversed (table 17) Methods and Materials Location Evans Experimental Farm, south Logan Treatments We established the same four treatments as in 1960 except that in the treatment enclosing honey bees intermit~ently, the bees were present 1 day in 4 instead of 1 day in 3. In addition, the eight plots caged with bees were divided into halves, the south half with 16-mesh cheesecloth sewed to it. The halves receiving shade could not be randomized because of the angle of the sun. Plot layout and onion lines The plot layout was the same as in The male-sterile lines were B-5546A, B-2264A, and U A, and the pollen parent line was U B. Lines B-2264A and B-5546A were the same as used in The pollen parent and line U A had no counterpart in the previous studies. Planting, irrigation and insect control The onion bulbs were set out on April 17. Irrigation water being scarce, the plants were only watered once each in May and June and not at all in July and August. Onion maggots and onion thrips were controlled before flowering began, as in previous years. ~feasurements 1. Readings of temperatures at flower height in the shaded and un- 29

33 shaded halves of the caged plots were made by the same methods used previously. 2. Nectar sugar concentrations were measured six times in each plot. 3. Counts on insects and pollination efficiency ratings were also taken six times during the season on the open and modified-open plots. The same number of population counts of honey bees were made in the plots enclosing honey bees, with separate counts being made in the shaded and unshaded halves. 4. The percentage of germination after 14 days was measured on the cleaned seed by the Vegetable Seed Investigations Laboratory, Agriculrural Research Service, Beltsville, Maryland. Conditions Affecting the Experiments Soil and soil moisture The soil was classified as Nibley silt loam. Soil conditions were drier than in previous years, especially late ih the season. Plant condition The onion bulbs were in good condition, and the plant growth was generally uniform from plot to plot. Since the plants were only irrigated twice, they did not grow as much as in previous years, but the difference was not obvious. Harmful insects Populations of harmful insects were held at a low level with insecticides applied before the onions began flowering. In July, with other crops drying up from lack of water, thrips and lygus bugs built up on the onions. The increase of these insects may have affected the seed yields in the open and modified-open plots. ~Veather and atmospheric conditions The weather was generally warm and clear. The average maximum temperature of 89.5 in a weather shelter was 3 degrees lower than in The average of four maximum temperature readings at flower height in the unshaded halves of the caged plot was 91.5 F., which was 4 degrees higher than that in the shaded halves. This difference was 30

34 significant at the I-percent level of probability. The average relative humidity was 12 percent, a few inches above the ground in the plots, as compared to 16 percent in Nectar sugar concentration The data on percent sugar in the nectar obtained from the contents of the honey stomachs of four nectar collectors per plot taken six times during the season can be summarized as follows: Open plots, 38.3; modified plots, 43.3; bees continuously, 26.8; bees intermittently, Except for the two uncaged treatments, all treatments were significantly different from each other at the I-percent level. Polinators in open and modified-open plots Tables 18 and 19 show the numbers and pollination indices of the insea visitors on the open and modified-open plots for each inbred line. The numbers in general were about the same as in 1959, several times as high as in 1958, and much lower than in Honey bees were about twice as abundant on the. open plots as on the modified-open plots, and other kinds of bees were nearly twice as abundant on the modifiedopen plots. Other insects generally showed ltttle in the way of differential response to the treatments. Differences in attractiveness between lines were less apparent than in previous years. In addition, for the first time, the pollen parents were as attractive as the most attractive male-sterile line. Figure 5. Onion pollination, enclosed plot with honey bees. 31

35 --~ Table Numbers and pollination indices of various kinds of pollinators per B-5546A B-5546A Pollinator No. Index No. Index Apis mellifera (pollen) Apis mellifera (nectar) Other bees All wasps Large Diptera Syritta pipiens Tiny Diptera Other insects All insects _...- Insects less honey bees Table 19. Numbers of various kinds of pollinators in the modified-open plots B-5546A B-5546A Pollinator No. Index ---- No. Index Apis mellifera (pollen Apis mellifera (nectar) Other bees All wasps Large Diptera Syritta pipiens Tiny Diptera Other insects All insects Insects less honey bees Table 20. Average number of insects and their efficiency per treatment per Logan, Utah, 1961 B-5546A B-5546A ----._ Treatment No. Index No. Index Open 19.9A * 19.5B 16.5C~- 17.7B Modified-open 14.8A 12.9B 19.8BC Bees 1 day in 4 (north side of cage not shaded) C 9.78 Bees 1 day in 4 (south side of cage shaded) 4.2B C 4.68 Bees continuously (north side of cage not shaded) A 52.5A 54.5A Bees continuously (south side of cage shaded) 35.7A 38.1AB 42.8AB 56.2A _.._ * Means not followed by the same letter are significantly different at the 1-percent level. 32

36 100 feet of row in the open plots, onion pollination experiment, Logan, Utah, Inbred line U B (pollen parent) B-2264A U A Average No. Index No. Index No. Index No. Index per 100 feet of row, onion pollination experiment, Logan, Utah, 1961 Inbred line --U-= B (pollen parent B-2264A U A Average No. Index No. Index No. Index No. Index ' i ~ feet of row on imbred lines bp treatment, onion pollination experiment, Inbred line U B (pollen parent) B-2264A U A Average No. Index No. Index No. Index No. Index B C 23.78C C B C A 87.8A 77.6A 79.8A 67.5A 75.0A 51.8A 69.2A 53.2A 70.9A 63.5A 64.7A 51.9A 55.2A 49.4A 57.0A 33

37 Pollinators in the plots enclosing bees Visitation to the onions was very low in the cages with bees enclosed 1 day our of 4 (an average of 6.5 per 100 feet of row) (table 20). In the cages with bees continuously enclosed, the populations were much higher (average 55.6 per 100 feet of row). In both treatments, the populations were higher in the unshaded than the shaded halves of the cages. Blasted florets Results The percent of blasted florets was generally lower than in 1960 (table 21). Some lines (for example, B-5546A and U B) had more blasting in the bee cages than in the open or modified-open plots, but in B-2264A the reverse was true in small measure. The shaded end of the plots caged with bees showed more blasting than the unshaded half, contrary to expectations. Seed yields Yields were lower than in 1959 and differences among treatments and among inbred lines were the smallest of the 4 years of study. Unlike 1959 and 1960, the plots with honey bees continuously enclosed had yields equivalent to this in the open and modified-open plots. The plots with honey bees enclosed 1 day of 4 (very light population) had significantly lower yields. In both honey bee treatments, the extra shading on the south halves of the cages resulted in increased yields (table 22). Seed germination The average 14-day germination of seeds from the four basic treatments ranged from 89.5 percent in the plots caged with honey bees continuously to 94.4 percent in the modified-open plots. Only the latter figure was significantly different from the others (table 23). There were significant differences in the seed germination of the inbred lines and these tended to be consistent from treatment to treatment. The inbred with the lowest germination was B-5546A (83.5 percent) and that with the highest was B-2264A (97.1 percent). Correlations When eight variables were matched with the seed yields, the follow- 34

38 ing were significantly positively correlated with the yields within treatment (table 24): (1) Pollination index (all treatments), (2) insect visitation (open and modified -open), (3) 14-da y germination (all treatments), (4) pollen collecting honey bees (open, bees enclosed continuously and intermittently), (5) nectar-collecting honey bees (open and modified-open), (6) Syritta pipiens (open), and (7) Eristalis tenax (open). Blasted florets (all except modified-open) were significantly negatively correlated on the same basis. When matched with yields within lines, the following correlations were significant (table 25): Pollination index (three lines negative), (2) insect visitation (same three iines negative), (3) blasted florets (all lines negative), (4) 14-day germination (two lines negative and one line positive), (5) pollen-collecting honey bees (two lines negative and one line positive), (6) nectar-collecting honey bees (all lines negative), (7) Syritta pipiens (two lines negative and one line positive, and one line with a negative correlation from one row (bulb source) and positive from another (the two rows from different bulb sources». In a correlation between pollination index and blasted florets, there was a significant negative correlation within all treatments except modified-open and a positive correlation within two lines. Figure 6. Onion pollinator, nectar-collecting honey bee. 35

39 Table 21. Percent of florets blasted in center of umbel by pollin ation level and inbred line, onion pollination experiment, Logan, Utah, 1961 Bees caged 1 day In 4 Bees caged continuously Inbred line Open Modified-open Light Shade Light Shade Average A C* 3.52 C 0.0 D 0.0 D 0.0 D 0.0 D A 0.0 D 0.0 D 0.0 D D 7.30 C A 0.0 D 0.0 D A8 0.0 D A 6.26 U A 0.0 D 0.0 D 0.0 D 0.0 D 0.0 D 0.0 D 0.0 U D 0.0 D 2.81 C A8C 0.0 D 9.63 A8C 3.96 (pollen parent) Average \.)) 0\ * Means not followed by the same letter are significantly diffe rent at the 1 percent level. Table 22. Seed yields in pounds per acre by inbred line and pollination level, onion pollination experiment, Logan, Utah, 1961 Bees caged 1 day in 4 Bees caged continuously Inbred line Open Modified-open Light Shade Light Shade Average A 214 A* 309 A A A 334 A 339 A A 407 A A 392 A. 442 A A 430 A 344 U A 243 A 269 A A 333 A 236 U A 275 A A 222 A 290 A 236 (pollen parent) Average * Means not followed by the same letter are significantly different at the 1 percent level.

40 Table 23. Germination (14 days) percentages of seed produced under levels of pollination, onion pollination experiment, Logan, Utah, 1961 U B B-5546A B-5546A (pollen parent) B-2264A U A Average Bees intermittently C* B A A B B Bees continuously C B A A C B Open C A A A B B Mod ified-open B A A A B A Average * Means followed by the same letter are not significantly dif ;erent from each other. I.).) ""-J Table 24. Correlations between variables obtained from the analysis by pollination levels, onion pollination experiment, Logan, Utah, 1961 Variables Index X seed yields Insects X seed yields Arc sine blast X seed yields Germination (14 days) X seed yields Honey bees (pollen) X seed yields Honey bees (nectar) X seed yields Syritta pipiens X seed yields Eristalis tenax X seed yields Index X blasting * Correlation coefficients significant at the 5-percent level. Correlation coefficients significant at the 1-percent level. Open.712**.501 * * -.245*.556* *.587* *.259* *.530* *.543* * -.283* Correlation coefficients Modified- Bees Beees open intermittent continuous.447* *.322* *.373**.317* * * * -.705**.613**.695* *.717** * *.502**.372* * * * -.405**

41 \,)) 00 Table 25. Correlations between variables obtained from the an alysis by inbred onion lines, onion pollination experiment, Logan, Utah, 1961 Inbred line U B Variables B-5546A B-5546A (pollen parent) B-2264A U A Index X seed yields -.608* * -.520* ~584* *.040 Insects X seed yields -.310* * -.290* **.021 Blast X seed yields -.738* * -.759** -.398* * -.418** -.313* * Germination (14-day) X seed yields * * -.423* * * * Honey bees (pollen) X seed yields * * -.386* *.365* *.152 Honey bees (nectar) X seed yields -.766** -.727** -.412** -.638* * -.305* * Syritta pipiens X seed yields -.857* * * *.572* * -.640* *.037 Eristalis tenax X seed yields -.318**.513**.553* * -.586* *.592* * Total index X blasting.476* * ** * Correlation coefficients significant at the 5-percent level. Correlation coefficients significant at the 1-percent level. Table 26. Numbers of various kinds of pollinators per 100 feet of row on open and modified-open plots, onion pollination experiment, Logan, Utah, Species Open Open Modified-open Open Modified-open Open Modified-open Apis mellifera (pollen) ~.----" Apis mellifera (nectar) 5.9* 92.3* Other bees All wasps Large Diptera Syritta pipiens Tiny Diptera Other insects All insects Insects less honey bees * Total honey bees for 1958 and 1959 ~---~-~---~,, ~...

42 DISCUSSION AND CONCLUSIONS The experiments from 1958 to 1%1 were successful in terms of achieving significant differences between treatments in insect visitation and seed yields. One treatment, that excluding all insects except tiny flies, resulted in no seed and was not used after The modified-open plots ( ), which allowed insects to come and go but subjected the plants to other aspects of the cage effect, reduced insect visitation somewhat and, hence, was not an ideal control treatment for the cage effect. This modified-open treatment reduced overall insect visitation about 20 percent to 50 percent during the 3 years (table 26). In spite of the lower insect visitation to the modified-open plots, seed yields were higher in these plots than in the open plots all 3 years. This indicated that the shading effect (possibly coupled with reduction in air movement) had a beneficial effect on yields. It also could indicate that reducing the insect visitation below that in the open plots tended to increase yields. However, insect visitation was positively correlated with seed yields on individual lines in 1959, even when the reverse influence of shading on populations was included by con~idering both open and modified-open plots together. The regression coefficient indicated that about 40 percent of the variation in yields was accounted for by differences in insect VIsitation. Thus, it would appear that, although insect visitation levels were important, the environmental condition brought about by the cages was more important within the ranges of insect visitation encountered. In a comparison of yields and pollination indices from year to year on the open and modified-open plots (table 27), it can be seen that in 1958 and 1961, these plots had the lowest yields and the lowest pollination indices. However, within the 2-year groups (198, 1961 and 1959, 1960), the yields and pollination indices were inversely related. Possible explanations are: (1) in 1961, irrigation water was severely limited, thus limiting plant capacity, and (2) in 1960, the pollination indices were probably overestimated as a result of overestimating the value of the tiny flies that made up a high proportion of the insect visitors that year. The results obtained from the enclosed plots with honey bees do not bear out the positive association between insect visitation and seed yields (table 28). Each year the plots with honey bees continuously enclosed had higher pollination indices than the modified-open plots, but they had lower yields (except in 1961 on the halves of the plots receiving extra shading). In 1960, reducing the exposure of the plots to honey bees to 1 day in 3 39

43 Table 27. Comparison of yields and pollination indices by yea.r on open and modified-open plots, onion pollination experiment, Logan, Utah, Open plots Modified-open plots Yield Pollination index Yield Pollination index Year Ibs/acre per 100 ft. of row Ibs/acre per 100 ft of row * * 1961** * Including indices from high percentage of tiny flies (probably overrated). * * Only two irrigations (adequate in previous years). Table 28. Comparisons of pollination indices and yields in open and caged plots by treatment and year, onion pollination experiment, Logan, Utah Pollination index Yield Year Treatment (per 100 ft of row) Ibs/acre 1958 Open Honey bees continuously 228* Open Modified-open Honey bees continuously 450* Open Modified-open Honey bees continuously Honey bees 1 day in Open Modified-open Honey bees continuously unshaded Honey bees continuously shaded Honey bees 1 day in 4 unshaded Honey bees 1 day in 4 - shaded * Based upon population estimates and measured percentages of pollen and nectar collectors. resulted in a three-fold reduction in VIsItation even on the days of exposure. In spite of this there was a seed yield increase. However, in 1961, when the exposure was further reduced to 1 day in 4 and there was a 6-fold population reduction on the day of exposure, yields were reduced by about half. Apparently, the caged honey bees had a harmful effect on pollination superimposed on their necessary basic role as pollinators. Probably, when the visitation was extremely low, as in the plots with bees present 1 day out of 4 greater visitation would have increased yields. On the other hand, increased visitation in the cages with honey bees above an index of 100 per 100 feet of row appeared to reduce yields. The greatest reduction occurred in 1960 when the yields were moderate in the modified-open plots but poor in the caged plots with bees. Since there were a number of very 40

44 hot days in 1960 (above 100 F.), it seems likely that excessive heat compounded the harmful effect of the bees. Further evidence for an interaction between heat and honey bee activity in the cages was provided by the blasting that occurred in the centers of many flower heads (table 16). Blasted florets were especially prevalent in 1960 in the cages enclosing bees. They were more prevalent in the cages with bees continuously present than in the ones in which visitation was intermittent (and light). The possibility of blasting being caused by eriophyid mites cannot be ruled out completely since we made no pertinent observations. However, it would appear that, if eriophyids were involved, their effect was greatest under conditions of high temperatures in cages with honey bees. Some studies were undertaken to find out what the honey bees in the cages were doing to depress yields. Physical injury to the stigmatic surface and stripping of pollen from the stigma were not observed but may have occurred.. There is no evidence that honey bees were harmful when not confined to cages. For example, honey bees were the dominant pollinators in the modified open plots in 1959 when the yields on these plots were the highest we obtained in the 4 years of our experiments. The high yields generally obtained in 1959 probably resulted from a combination of good weather (warm and sunny, but not hot), good Figure 7. Onion pollinator Sphex ichneumoneus. 41